Repository navigation
Expand file tree
/
Copy pathindex.html
More file actions
629 lines (609 loc) · 54 KB
/
Copy pathindex.html
File metadata and controls
629 lines (609 loc) · 54 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta name="description" content="DeepAgentLabs builds open standards and tools to observe, optimize, and stress-test production AI systems.">
<meta name="theme-color" content="#07110e">
<meta property="og:title" content="DeepAgentLabs — AI operations, built in the open">
<meta property="og:description" content="A specification-first ecosystem for observable, efficient, and resilient AI systems.">
<meta property="og:type" content="website">
<meta property="og:url" content="https://deepagentlabs.io/">
<meta property="og:image" content="https://deepagentlabs.io/assets/img/ecosystem.png">
<meta name="twitter:card" content="summary_large_image">
<meta name="twitter:image" content="https://deepagentlabs.io/assets/img/ecosystem.png">
<link rel="canonical" href="https://deepagentlabs.io/">
<title>DeepAgentLabs — Open AI Operations</title>
<link rel="stylesheet" href="assets/css/styles.css">
<link rel="stylesheet" href="assets/css/architecture.css">
<link rel="stylesheet" href="assets/css/enhance.css">
<link rel="stylesheet" href="assets/css/adoption.css">
<link rel="stylesheet" href="assets/css/roadmap.css">
<link rel="stylesheet" href="assets/css/nav.css">
<link rel="stylesheet" href="assets/css/theme.css">
<link rel="stylesheet" href="assets/css/background.css">
<link rel="stylesheet" href="assets/css/copy.css">
<link rel="stylesheet" href="assets/css/tools.css">
<link rel="stylesheet" href="assets/css/architecture-theme.css">
<link rel="stylesheet" href="assets/css/adoption-card.css">
<link rel="stylesheet" href="assets/css/polish.css">
<link rel="stylesheet" href="assets/css/mobile.css">
<script src="assets/js/site.js" defer></script>
<script src="assets/js/architecture.js" defer></script>
<script src="assets/js/enhance.js" defer></script>
<script src="assets/js/adoption.js" defer></script>
<script src="assets/js/roadmap.js" defer></script>
<script src="assets/js/usecases-data.js" defer></script>
<script src="assets/js/usecases.js" defer></script>
<script src="assets/js/nav.js" defer></script>
<script src="assets/js/background.js" defer></script>
<script src="assets/js/copy.js" defer></script>
<script src="assets/js/tools.js" defer></script>
<script src="assets/js/architecture-compact.js" defer></script>
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.5.2/css/fontawesome.min.css">
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.5.2/css/brands.min.css">
<link rel="stylesheet" href="assets/css/community.css">
<link rel="stylesheet" href="assets/css/footer.css">
<link rel="stylesheet" href="assets/css/beam.css">
<script src="assets/js/beam.js" defer></script>
</head>
<body>
<a class="skip-link" href="#main">Skip to content</a>
<header class="site-header">
<a class="brand" href="#top" aria-label="DeepAgentLabs home">
<span class="brand-logo" aria-hidden="true"></span>
<span>DeepAgentLabs</span>
</a>
<button class="nav-toggle" type="button" aria-expanded="false" aria-controls="site-nav">Menu</button>
<nav id="site-nav" aria-label="Primary navigation">
<div class="nav-item has-mega">
<button class="nav-trigger" type="button" aria-expanded="false" aria-controls="mega-platform">Platform</button>
<div class="mega mega-compact mega-narrow" id="mega-platform">
<div class="mega-main">
<p class="mega-label">On this site</p>
<ul class="mega-grid">
<li><a class="mega-link mega-link-plain" href="#loop"><span class="mega-text"><strong>The operating loop <i aria-hidden="true">→</i></strong><small>Define, observe, stress, evaluate, improve.</small></span></a></li>
<li><a class="mega-link mega-link-plain" href="#dashboards"><span class="mega-text"><strong>Dashboards <i aria-hidden="true">→</i></strong><small>What each product shows you.</small></span></a></li>
<li><a class="mega-link mega-link-plain" href="#why"><span class="mega-text"><strong>Why DeepAgentLabs <i aria-hidden="true">→</i></strong><small>How we differ from dashboards and SDK evals.</small></span></a></li>
<li><a class="mega-link mega-link-plain" href="#community"><span class="mega-text"><strong>Community <i aria-hidden="true">→</i></strong><small>Discord, LinkedIn, Instagram and GitHub.</small></span></a></li>
</ul>
</div>
<a class="mega-side" href="roadmap.html">
<span class="mega-label">Roadmap</span>
<strong>Step 1 of 8: v0.1 in review.</strong>
<span class="mega-side-copy">See where the specification is headed, phase by phase.</span>
<span class="mega-side-cta">View roadmap <svg viewBox="0 0 24 24" aria-hidden="true"><path d="M5 12h14M13 6l6 6-6 6"/></svg></span>
</a>
</div>
</div>
<div class="nav-item has-mega">
<button class="nav-trigger" type="button" aria-expanded="false" aria-controls="mega-products">Products</button>
<div class="mega" id="mega-products">
<div class="mega-main">
<p class="mega-label">Products</p>
<ul class="mega-grid">
<li><a class="mega-link" href="/agenticlens"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><circle cx="11" cy="11" r="6"/><path d="m20 20-4.5-4.5"/><circle cx="11" cy="11" r="2"/></svg></span><span class="mega-text"><strong>AgenticLens <i aria-hidden="true">→</i></strong><small>Observe — OTLP traces, cost and latency on live traffic.</small></span></a></li>
<li><a class="mega-link" href="/agentic-chaos"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M13 2 5 13h6l-1 9 8-11h-6l1-9z"/></svg></span><span class="mega-text"><strong>Agentic Chaos <i aria-hidden="true">→</i></strong><small>Stress — fault injection and drift testing across frameworks.</small></span></a></li>
<li><a class="mega-link" href="/agentic-sidecar"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M12 3 4 6v6c0 4.5 3.4 8 8 9 4.6-1 8-4.5 8-9V6l-8-3z"/><path d="m9 12 2 2 4-4"/></svg></span><span class="mega-text"><strong>Agentic Sidecar <i aria-hidden="true">→</i></strong><small>Govern — allow, warn or block actions against intent and policy.</small></span></a></li>
<li><a class="mega-link" href="/agentic-evals"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M4 19V9M10 19V5M16 19v-7M22 19H2"/></svg></span><span class="mega-text"><strong>Agentic Evals <i aria-hidden="true">→</i></strong><small>Evaluate — deterministic and LLM-judge scoring behind a CI gate.</small></span></a></li>
<li><a class="mega-link" href="/mcp"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M9 7V3M15 7V3M7 7h10v4a5 5 0 0 1-10 0V7zM12 16v5"/></svg></span><span class="mega-text"><strong>MCP Server <i aria-hidden="true">→</i></strong><small>Connect — every tool exposed through one MCP interface.</small></span></a></li>
<li><a class="mega-link" href="/control-tower"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><rect x="3" y="3" width="7" height="7"/><rect x="14" y="3" width="7" height="7"/><rect x="3" y="14" width="7" height="7"/><rect x="14" y="14" width="7" height="7"/></svg></span><span class="mega-text"><strong>Control Tower <i aria-hidden="true">→</i></strong><small>Operate — registry and console for every stage.</small></span></a></li>
</ul>
</div>
<a class="mega-side" href="https://github.com/DeepAgentLabs/ai-operations-spec">
<span class="mega-label">The shared contract</span>
<strong>AI Operations Specification</strong>
<span class="mega-side-copy">One vendor-neutral spec every product implements — so runs, steps and evaluations mean the same thing everywhere.</span>
<span class="mega-side-cta">Read the spec <svg viewBox="0 0 24 24" aria-hidden="true"><path d="M5 12h14M13 6l6 6-6 6"/></svg></span>
</a>
</div>
</div>
<div class="nav-item has-mega">
<button class="nav-trigger" type="button" aria-expanded="false" aria-controls="mega-docs">Docs</button>
<div class="mega" id="mega-docs">
<div class="mega-main">
<p class="mega-label">Documentation</p>
<ul class="mega-grid">
<li><a class="mega-link" href="/agenticlens/docs"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M6 3h8l4 4v14H6z"/><path d="M14 3v4h4M9 12h6M9 16h6"/></svg></span><span class="mega-text"><strong>AgenticLens docs <i aria-hidden="true">→</i></strong><small><code>pip install agenticlens</code></small></span></a></li>
<li><a class="mega-link" href="/agentic-chaos/docs"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M6 3h8l4 4v14H6z"/><path d="M14 3v4h4M9 12h6M9 16h6"/></svg></span><span class="mega-text"><strong>Agentic Chaos docs <i aria-hidden="true">→</i></strong><small><code>pip install agentic-chaos</code></small></span></a></li>
<li><a class="mega-link" href="/agentic-sidecar/docs"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M6 3h8l4 4v14H6z"/><path d="M14 3v4h4M9 12h6M9 16h6"/></svg></span><span class="mega-text"><strong>Agentic Sidecar docs <i aria-hidden="true">→</i></strong><small><code>pip install agentic-sidecar</code></small></span></a></li>
<li><a class="mega-link" href="/agentic-evals/docs"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M6 3h8l4 4v14H6z"/><path d="M14 3v4h4M9 12h6M9 16h6"/></svg></span><span class="mega-text"><strong>Agentic Evals docs <i aria-hidden="true">→</i></strong><small><code>pip install agentic-evals</code></small></span></a></li>
<li><a class="mega-link" href="/mcp/docs"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M6 3h8l4 4v14H6z"/><path d="M14 3v4h4M9 12h6M9 16h6"/></svg></span><span class="mega-text"><strong>MCP Server docs <i aria-hidden="true">→</i></strong><small>Tools, resources and client setup.</small></span></a></li>
<li><a class="mega-link" href="/control-tower/docs"><span class="mega-icon"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M6 3h8l4 4v14H6z"/><path d="M14 3v4h4M9 12h6M9 16h6"/></svg></span><span class="mega-text"><strong>Control Tower docs <i aria-hidden="true">→</i></strong><small><code>pip install agenticops-control-tower</code></small></span></a></li>
</ul>
</div>
<div class="mega-side mega-side-static">
<span class="mega-label">Start here</span>
<strong>Up and running in a minute.</strong>
<code class="mega-code">pip install agenticlens</code>
<span class="mega-side-copy">Instrument one agent, see its first trace, then add evals and chaos when you're ready.</span>
<span class="mega-label mega-pypi-label">On PyPI</span>
<div class="mega-pypi"><a href="https://pypi.org/project/agenticlens/" target="_blank" rel="noopener noreferrer">AgenticLens <span aria-hidden="true">↗</span></a><a href="https://pypi.org/project/agentic-chaos/" target="_blank" rel="noopener noreferrer">Chaos <span aria-hidden="true">↗</span></a><a href="https://pypi.org/project/agentic-evals/" target="_blank" rel="noopener noreferrer">Evals <span aria-hidden="true">↗</span></a><a href="https://pypi.org/project/agentic-sidecar/" target="_blank" rel="noopener noreferrer">Sidecar <span aria-hidden="true">↗</span></a><a href="https://pypi.org/project/deep-agentic-core-mcp/" target="_blank" rel="noopener noreferrer">MCP <span aria-hidden="true">↗</span></a><a href="https://pypi.org/project/agenticops-control-tower/" target="_blank" rel="noopener noreferrer">Control Tower <span aria-hidden="true">↗</span></a></div>
<a class="mega-side-cta" href="/agenticlens/docs">AgenticLens quickstart <svg viewBox="0 0 24 24" aria-hidden="true"><path d="M5 12h14M13 6l6 6-6 6"/></svg></a>
</div>
</div>
</div>
<!-- USE CASES MENU: content comes from assets/js/usecases-data.js (edit it there, not here).
assets/js/usecases.js fills the [data-uc-list] and [data-uc-most-asked] slots below. -->
<div class="nav-item has-mega">
<button class="nav-trigger" type="button" aria-expanded="false" aria-controls="mega-usecases">Use cases</button>
<div class="mega mega-usecases" id="mega-usecases">
<div class="mega-main">
<p class="mega-label">Use cases</p>
<ul class="mega-grid" data-uc-list></ul>
</div>
<div class="mega-side mega-side-static">
<span class="mega-label">Start here</span>
<strong>See it on a real agent.</strong>
<span class="mega-side-copy">Each use case walks through the problem, the setup and the result, step by step.</span>
<span class="mega-label mega-pypi-label">Most asked</span>
<div class="mega-pypi" data-uc-most-asked></div>
<a class="mega-side-cta" href="https://discord.gg/fEqPFWgcDG" target="_blank" rel="noopener noreferrer">Share your use case <svg viewBox="0 0 24 24" aria-hidden="true"><path d="M5 12h14M13 6l6 6-6 6"/></svg></a>
</div>
</div>
</div>
<!-- /USE CASES MENU -->
<a href="#architecture">Architecture</a>
<a href="#faq">FAQ</a>
<a href="https://github.com/DeepAgentLabs/ai-operations-spec">Specification</a>
<a href="roadmap.html">Roadmap</a>
<a href="https://mcp.deepagentlabs.io" target="_blank" rel="noopener noreferrer">Remote MCP</a>
<a class="nav-cta" href="https://github.com/DeepAgentLabs">GitHub <span aria-hidden="true">↗</span></a>
</nav>
</header>
<main id="main">
<section class="hero" id="top">
<div class="hero-copy">
<p class="eyebrow"><span class="status-dot"></span> Building in the open</p>
<h1>AI operations need a <em>shared language.</em></h1>
<p class="hero-lede">DeepAgentLabs is a specification-first ecosystem for understanding how AI systems run, what they cost, and how they fail. We write one open standard for operating AI agents and build free tools on top of it.</p>
<div class="hero-actions">
<a class="button button-primary" href="https://github.com/DeepAgentLabs/ai-operations-spec">Explore the specification</a>
<a class="button button-secondary" href="#dashboards">See the ecosystem</a>
</div>
<dl class="hero-proof hero-about" aria-label="About DeepAgentLabs">
<div><dt>Who we are</dt><dd>An open-source AI research and learning community.</dd></div>
<div><dt>What we build</dt><dd>One open spec and six free tools for AI agents.</dd></div>
<div><dt>Who it's for</dt><dd>Anyone running agents. Any framework, runs locally.</dd></div>
</dl>
</div>
<div class="system-card" aria-label="DeepAgentLabs ecosystem model">
<div class="system-card-top"><span>ECOSYSTEM / 01</span><span class="pulse"><i class="live-dot" aria-hidden="true"></i>Live</span></div>
<img class="hero-diagram" src="assets/img/ecosystem.png" alt="DeepAgentLabs reference architecture: ai-operations-spec as the shared contract connecting AgenticLens, Agentic Evals, Agentic Chaos, Agentic Sidecar, AgenticOps Control Tower, and the MCP server.">
<div class="system-footer"><span>One model</span><span>Multiple implementations</span></div>
</div>
</section>
<section class="loop section-shell" id="loop">
<div class="section-heading">
<div><p class="section-label">The operating loop</p><h2>One loop,<br><span class="hl">five stages.</span></h2></div>
<p>Define the contract once, then observe, stress-test, evaluate, and improve every agent against it.</p>
</div>
<ol class="loop-track">
<li class="loop-step"><span class="loop-index">01</span><strong>Define</strong><span class="loop-pkg">ai-operations-spec</span><p>The vendor-neutral contract for runs, steps, and evaluations every tool here implements.</p></li>
<li class="loop-step"><span class="loop-index">02</span><strong>Observe</strong><span class="loop-pkg">agenticlens</span><p>OTLP trace ingest, cost and latency tracking on live traffic.</p></li>
<li class="loop-step"><span class="loop-index">03</span><strong>Stress</strong><span class="loop-pkg">agentic-chaos</span><p>Fault injection and topology-aware drift testing across agent frameworks.</p></li>
<li class="loop-step"><span class="loop-index">04</span><strong>Evaluate</strong><span class="loop-pkg">agentic-evals</span><p>Deterministic and LLM-as-judge scoring behind a pass/fail CI gate.</p></li>
<li class="loop-step"><span class="loop-index">05</span><strong>Improve</strong><span class="loop-pkg">control-tower</span><p>Registry and console tying every stage into one operating surface.</p></li>
</ol>
</section>
<section class="architecture section-shell" id="architecture">
<div class="section-heading">
<div><p class="section-label">Architecture</p><h2>The full <span class="hl">capability map.</span></h2></div>
<p>Every capability each package exposes, and how a request moves from a user through the Control Tower down to the packages doing the work.</p>
</div>
<figure class="architecture-figure">
<div class="arch-legend" aria-label="Relationship status legend">
<span><i class="arch-swatch arch-sw-warn" aria-hidden="true"></i>Current integration</span>
<span><i class="arch-swatch arch-sw-plan" aria-hidden="true"></i>Planned integration</span>
<span><i class="arch-swatch arch-sw-context" aria-hidden="true"></i>Architecture context</span>
</div>
<div class="arch-toolbar" aria-label="Architecture controls">
<button id="arch-trace" type="button" aria-pressed="false"><span aria-hidden="true">▶</span> Trace request</button>
<button id="arch-reset" type="button"><span aria-hidden="true">↺</span> Reset</button>
</div>
<div id="arch-infographic" class="arch-infographic" role="group" aria-labelledby="arch-title arch-description">
<h3 class="sr-only" id="arch-title">DeepAgentLabs interactive architecture infographic</h3>
<p class="sr-only" id="arch-description">Explore the full ecosystem infographic. Select components, capabilities, frameworks, or outputs for details. Use arrow keys to move among major components, then Enter or Space to select.</p>
<div id="arch-artwork" class="arch-artwork"></div>
<aside class="arch-panel" aria-label="Architecture details" aria-live="polite" aria-atomic="true">
<div id="arch-panel-empty">
<p class="arch-panel-eyebrow">Component details</p>
<p class="arch-panel-empty">Select a component, capability, framework, or export to inspect its role and integration status.</p>
</div>
<div id="arch-panel-detail" hidden>
<p class="arch-panel-eyebrow" id="arch-pd-kind"></p>
<h3 id="arch-pd-name"></h3>
<span class="arch-ver" id="arch-pd-version"></span>
<p class="arch-desc" id="arch-pd-desc"></p>
<h4 class="arch-edges-title">Connected components</h4>
<div id="arch-pd-edges"></div>
</div>
<div id="arch-trace-status" class="arch-trace-status" aria-live="polite" hidden></div>
</aside>
</div>
</figure>
</section>
<section class="dashboards section-shell" id="dashboards">
<div class="section-heading">
<div><p class="section-label">See it in action</p><h2>What each framework<br><span class="hl">actually does.</span></h2></div>
<p>Seven frameworks, one contract. Pick one to see what it does and its sample output. These are illustrative examples, not live telemetry.</p>
</div>
<div class="tl" data-tl>
<div class="tl-tabs" role="tablist" aria-label="DeepAgentLabs frameworks">
<button class="tl-tab" type="button" role="tab" id="tl-tab-spec" aria-controls="tl-panel-spec" aria-selected="true" tabindex="0"><span class="tl-verb">01 · Define</span><strong>AI Operations Spec</strong><i class="tl-progress" aria-hidden="true"></i></button>
<button class="tl-tab" type="button" role="tab" id="tl-tab-lens" aria-controls="tl-panel-lens" aria-selected="false" tabindex="-1"><span class="tl-verb">02 · Observe</span><strong>AgenticLens</strong><i class="tl-progress" aria-hidden="true"></i></button>
<button class="tl-tab" type="button" role="tab" id="tl-tab-evals" aria-controls="tl-panel-evals" aria-selected="false" tabindex="-1"><span class="tl-verb">03 · Evaluate</span><strong>Agentic Evals</strong><i class="tl-progress" aria-hidden="true"></i></button>
<button class="tl-tab" type="button" role="tab" id="tl-tab-chaos" aria-controls="tl-panel-chaos" aria-selected="false" tabindex="-1"><span class="tl-verb">04 · Stress</span><strong>Agentic Chaos</strong><i class="tl-progress" aria-hidden="true"></i></button>
<button class="tl-tab" type="button" role="tab" id="tl-tab-sidecar" aria-controls="tl-panel-sidecar" aria-selected="false" tabindex="-1"><span class="tl-verb">05 · Govern</span><strong>Agentic Sidecar</strong><i class="tl-progress" aria-hidden="true"></i></button>
<button class="tl-tab" type="button" role="tab" id="tl-tab-tower" aria-controls="tl-panel-tower" aria-selected="false" tabindex="-1"><span class="tl-verb">06 · Operate</span><strong>Control Tower</strong><i class="tl-progress" aria-hidden="true"></i></button>
<button class="tl-tab" type="button" role="tab" id="tl-tab-mcp" aria-controls="tl-panel-mcp" aria-selected="false" tabindex="-1"><span class="tl-verb">07 · Connect</span><strong>DeepAgent MCP</strong><i class="tl-progress" aria-hidden="true"></i></button>
</div>
<div class="tl-stage">
<article class="dash-card tl-panel" role="tabpanel" id="tl-panel-spec" aria-labelledby="tl-tab-spec">
<div class="dash-head"><strong>AI Operations Specification</strong><span class="dash-tag">v0.4 draft</span></div>
<p class="dash-desc">The vendor-neutral contract for runs, steps, model/tool calls, evaluations, and incidents — the shared vocabulary every tool on this page implements, not an SDK or a runtime itself.</p>
<p class="dash-caps-title">What it defines</p>
<ul class="dash-caps">
<li>Runs, steps & incidents schema</li>
<li>Model & tool call vocabulary</li>
<li>Evaluation & safety signals</li>
<li>JSON Schema artifacts, v0.1–v0.4</li>
</ul>
<ul class="dash-check" aria-hidden="true">
<li><span class="dot dot-good"></span>schema_valid</li>
<li><span class="dot dot-good"></span>semantic_valid</li>
<li><span class="dot dot-good"></span>aligned: true</li>
</ul>
<p class="sr-only">Conformance check example: schema valid, semantic valid, aligned true.</p>
<p class="dash-caption">One shared contract every framework below can validate against.</p>
<div class="tl-foot"><span></span><a class="text-link" href="https://github.com/DeepAgentLabs/ai-operations-spec">Read the standard <span aria-hidden="true">↗</span></a></div>
</article>
<article class="dash-card tl-panel" role="tabpanel" id="tl-panel-lens" aria-labelledby="tl-tab-lens" hidden>
<div class="dash-head"><strong>AgenticLens</strong><span class="dash-tag">workflow profile</span></div>
<p class="dash-desc">OpenTelemetry-native observability for production AI: ingest OTLP traces, track cost and latency per run, and surface evidence-backed findings for where to cut waste.</p>
<p class="dash-caps-title">Capabilities</p>
<ul class="dash-caps">
<li>OTLP trace ingestion</li>
<li>Cost & token tracking</li>
<li>Latency (p50/p95) breakdown</li>
<li>Built-in evaluation engine</li>
<li>Pass/fail release gates</li>
<li>Cost-cutting recommendations</li>
</ul>
<div class="dash-stats" aria-hidden="true">
<div><strong>$0.041</strong><span>avg cost / run</span></div>
<div><strong>412ms</strong><span>p95 latency</span></div>
<div><strong>6</strong><span>findings</span></div>
</div>
<div class="dash-spark" aria-hidden="true">
<span style="height:78%"></span><span style="height:64%"></span><span style="height:52%"></span><span style="height:40%"></span><span style="height:29%"></span>
</div>
<p class="sr-only">Cost per run trending down across five recent runs, from $0.087 to $0.041, alongside six optimization findings.</p>
<p class="dash-caption">Cost trending down over five runs after acting on its own findings.</p>
<div class="tl-foot"><code class="tl-pip">pip install agenticlens</code><span class="tl-links"><a class="text-link tl-pypi" href="https://pypi.org/project/agenticlens/" target="_blank" rel="noopener noreferrer">PyPI <span aria-hidden="true">↗</span></a><a class="text-link" href="https://github.com/DeepAgentLabs/agenticlens">View AgenticLens <span aria-hidden="true">↗</span></a></span></div>
</article>
<article class="dash-card tl-panel" role="tabpanel" id="tl-panel-evals" aria-labelledby="tl-tab-evals" hidden>
<div class="dash-head"><strong>Agentic Evals</strong><span class="dash-tag">release gate</span></div>
<p class="dash-desc">Framework-agnostic scoring for any trace-shaped run — deterministic checks, rubric scorers, and LLM-as-judge evaluators, wired into a pass/fail gate you drop into CI.</p>
<p class="dash-caps-title">Capabilities</p>
<ul class="dash-caps">
<li>Deterministic scorers (exact/regex/JSON diff)</li>
<li>10 built-in LLM-judge rubrics</li>
<li>Trajectory & tool-call scorers</li>
<li>One-line <code>Eval()</code> quickstart API</li>
<li>Declarative TestSuite/TestCase API</li>
<li>CI release gate + CLI</li>
</ul>
<ul class="dash-gate" aria-hidden="true">
<li><span><span class="dot dot-good"></span>trajectory-quality ≥ 0.85</span><span class="val">0.91</span></li>
<li><span><span class="dot dot-good"></span>cost-per-run ≤ $0.05</span><span class="val">$0.031</span></li>
<li><span><span class="dot dot-warn"></span>hallucination-rate ≤ 2%</span><span class="val">1.8%</span></li>
</ul>
<p class="sr-only">Release gate: trajectory quality 0.91 passes an 0.85 threshold, cost per run $0.031 passes a $0.05 cap, hallucination rate 1.8% sits inside a 2% warning band.</p>
<p class="dash-caption">Three thresholds, one CI gate — the merge blocks until every run clears them.</p>
<div class="tl-foot"><code class="tl-pip">pip install agentic-evals</code><span class="tl-links"><a class="text-link tl-pypi" href="https://pypi.org/project/agentic-evals/" target="_blank" rel="noopener noreferrer">PyPI <span aria-hidden="true">↗</span></a><a class="text-link" href="https://github.com/DeepAgentLabs/agentic-evals">View Agentic Evals <span aria-hidden="true">↗</span></a></span></div>
</article>
<article class="dash-card tl-panel" role="tabpanel" id="tl-panel-chaos" aria-labelledby="tl-tab-chaos" hidden>
<div class="dash-head"><strong>Agentic Chaos</strong><span class="dash-tag">120 injected faults</span></div>
<p class="dash-desc">Fault injection and topology-aware drift testing across agent frameworks — provider failures, corrupted outputs, and tool errors, forced before production does it for you.</p>
<p class="dash-caps-title">Capabilities</p>
<ul class="dash-caps">
<li>Timeout & rate-limit injection</li>
<li>Silent output corruption</li>
<li>Tool & memory faults</li>
<li>Topology-aware drift detection</li>
<li>LangGraph fault wrapper</li>
<li>Standalone JSON chaos report</li>
</ul>
<div class="dash-bar" aria-hidden="true">
<span class="seg seg-good" style="width:68%"></span><span class="seg seg-warn" style="width:22%"></span><span class="seg seg-critical" style="width:10%"></span>
</div>
<ul class="dash-legend" aria-hidden="true">
<li><span class="dot dot-good"></span>Recovered · 68%</li>
<li><span class="dot dot-warn"></span>Degraded · 22%</li>
<li><span class="dot dot-critical"></span>Failed · 10%</li>
</ul>
<p class="sr-only">Outcome of 120 injected faults: 68% recovered, 22% degraded, 10% failed.</p>
<p class="dash-caption">One in ten injected faults broke the agent outright — before a user hit it.</p>
<div class="tl-foot"><code class="tl-pip">pip install agentic-chaos</code><span class="tl-links"><a class="text-link tl-pypi" href="https://pypi.org/project/agentic-chaos/" target="_blank" rel="noopener noreferrer">PyPI <span aria-hidden="true">↗</span></a><a class="text-link" href="https://github.com/DeepAgentLabs/agentic-chaos">View Agentic Chaos <span aria-hidden="true">↗</span></a></span></div>
</article>
<article class="dash-card tl-panel" role="tabpanel" id="tl-panel-sidecar" aria-labelledby="tl-tab-sidecar" hidden>
<div class="dash-head"><strong>Agentic Sidecar</strong><span class="dash-tag">500 decisions</span></div>
<p class="dash-desc">A real-time Decision Gate that watches an agent's next action against its declared intent — catching moves that are technically permitted but contextually wrong, not just disallowed content.</p>
<p class="dash-caps-title">Capabilities</p>
<ul class="dash-caps">
<li>Real-time Decision Gate</li>
<li>Policy Advisor + Risk Evaluator</li>
<li>Intent Guardian (drift detection)</li>
<li>Allow / Warn / Block outcomes</li>
<li>Observe mode & Govern mode</li>
<li>LangGraph adapter</li>
</ul>
<div class="dash-bar" aria-hidden="true">
<span class="seg seg-1" style="width:92.4%"></span><span class="seg seg-2" style="width:5.8%"></span><span class="seg seg-3" style="width:1.8%"></span>
</div>
<ul class="dash-legend" aria-hidden="true">
<li><span class="dot dot-1"></span>Allow · 462</li>
<li><span class="dot dot-2"></span>Warn · 29</li>
<li><span class="dot dot-3"></span>Block · 9</li>
</ul>
<p class="sr-only">Decision Gate outcomes across 500 actions: 462 allowed, 29 warned, 9 blocked as out of scope.</p>
<p class="dash-caption">Escalating intervention, not a failure rate — a block is the Sidecar working.</p>
<div class="tl-foot"><code class="tl-pip">pip install agentic-sidecar</code><span class="tl-links"><a class="text-link tl-pypi" href="https://pypi.org/project/agentic-sidecar/" target="_blank" rel="noopener noreferrer">PyPI <span aria-hidden="true">↗</span></a><a class="text-link" href="https://github.com/DeepAgentLabs/agentic-sidecar">View Agentic Sidecar <span aria-hidden="true">↗</span></a></span></div>
</article>
<article class="dash-card tl-panel" role="tabpanel" id="tl-panel-tower" aria-labelledby="tl-tab-tower" hidden>
<div class="dash-head"><strong>AgenticOps Control Tower</strong><span class="dash-tag">48 agents</span></div>
<p class="dash-desc">The fleet-wide registry and control plane: which agents exist, what's installed, which framework they run on, and what needs attention right now.</p>
<p class="dash-caps-title">Capabilities today</p>
<ul class="dash-caps">
<li>In-memory agent registry</li>
<li>Capability discovery</li>
<li>Heartbeat-based health</li>
<li>Optional HTTP API & CLI</li>
</ul>
<div class="dash-bar" aria-hidden="true">
<span class="seg seg-good" style="width:85.4%"></span><span class="seg seg-warn" style="width:10.4%"></span><span class="seg seg-critical" style="width:4.2%"></span>
</div>
<ul class="dash-legend" aria-hidden="true">
<li><span class="dot dot-good"></span>Healthy · 41</li>
<li><span class="dot dot-warn"></span>Degraded · 5</li>
<li><span class="dot dot-critical"></span>Unhealthy · 2</li>
</ul>
<p class="sr-only">Fleet status across 48 registered agents: 41 healthy, 5 degraded, 2 unhealthy.</p>
<p class="dash-caption">One registry for every agent's runtime, framework, and capability versions.</p>
<div class="tl-foot"><code class="tl-pip">pip install agenticops-control-tower</code><span class="tl-links"><a class="text-link tl-pypi" href="https://pypi.org/project/agenticops-control-tower/" target="_blank" rel="noopener noreferrer">PyPI <span aria-hidden="true">↗</span></a><a class="text-link" href="https://github.com/DeepAgentLabs/agenticops-control-tower">View Control Tower <span aria-hidden="true">↗</span></a></span></div>
</article>
<article class="dash-card tl-panel" role="tabpanel" id="tl-panel-mcp" aria-labelledby="tl-tab-mcp" hidden>
<div class="dash-head"><strong>DeepAgent MCP</strong><span class="dash-tag">live surface</span></div>
<p class="dash-desc">A unified MCP interface over the shared model — exposing AgenticLens, Agentic Chaos, Agentic Sidecar, and Spec validation as one set of tools any MCP client can call.</p>
<p class="dash-caps-title">Capabilities</p>
<ul class="dash-caps">
<li>14 callable tools</li>
<li>Wraps Lens, Chaos, Sidecar & Spec</li>
<li>2 shared resources</li>
<li>3 prompt templates</li>
<li>stdio transport, session state</li>
<li>One identity for any MCP client</li>
</ul>
<div class="dash-stats" aria-hidden="true">
<div><strong>14</strong><span>tools exposed</span></div>
<div><strong>1,284</strong><span>calls routed today</span></div>
<div><strong>4/5</strong><span>adapters connected</span></div>
</div>
<p class="sr-only">14 tools exposed, 1,284 calls routed today, 4 of 5 sibling adapters connected.</p>
<p class="dash-caption">One MCP-native surface over every framework above, degrading honestly when one's missing.</p>
<div class="tl-foot"><code class="tl-pip">pip install deep-agentic-core-mcp</code><span class="tl-links"><a class="text-link tl-pypi" href="https://pypi.org/project/deep-agentic-core-mcp/" target="_blank" rel="noopener noreferrer">PyPI <span aria-hidden="true">↗</span></a><a class="text-link" href="https://github.com/DeepAgentLabs/mcp-server">View the MCP server <span aria-hidden="true">↗</span></a></span></div>
</article>
</div>
</div>
</section>
<section class="adoption section-shell" id="adoption" aria-labelledby="adoption-title">
<div class="section-heading adoption-heading">
<div>
<p class="section-label">Ecosystem adoption</p>
<h2 id="adoption-title">The ecosystem <span class="hl">is moving.</span></h2>
<h3>See it in the numbers.</h3>
</div>
<p>Live PyPI download statistics across the DeepAgentLabs ecosystem.</p>
</div>
<div class="adp-card" data-adp>
<p class="adp-kicker"><span class="adp-live" aria-hidden="true"></span><span id="adoption-data-status" role="status" aria-live="polite">Loading PyPI statistics…</span></p>
<div class="adp-total">
<strong id="adp-total" aria-live="off">0</strong>
<span>downloads across the DeepAgentLabs ecosystem</span>
</div>
<ol class="adp-grid" id="adp-grid" aria-label="PyPI downloads per package"></ol>
<div class="adp-foot">
<p>PyPI totals can include automated and CI downloads. Pepy badge totals may be cached up to 12 hours.</p>
<a class="button button-primary" href="#dashboards">Explore the ecosystem <span aria-hidden="true">→</span></a>
</div>
</div>
</section>
<section class="roadmap section-shell" id="roadmap">
<div class="section-heading compact">
<div><p class="section-label">Roadmap</p><h2>Where the specification<br><span class="hl">is headed.</span></h2></div>
<a class="text-link" href="roadmap.html">See the full roadmap <span aria-hidden="true">→</span></a>
</div>
<div class="rm-rail" data-rm-animate aria-hidden="true"><span class="rm-rail-fill"></span></div>
<div class="roadmap-track rm-track">
<a class="phase active" href="roadmap.html#v0-1" data-status="review"><span class="rm-top"><span class="rm-ver">v0.1</span><span class="rm-status"><i></i>In review</span></span><strong>Core concepts</strong><p>Shared meaning for runs, steps, agents and tools.</p><b class="rm-arrow" aria-hidden="true">→</b></a>
<a class="phase" href="roadmap.html#v0-2" data-status="draft"><span class="rm-top"><span class="rm-ver">v0.2</span><span class="rm-status"><i></i>Exploratory</span></span><strong>Relationships</strong><p>How the pieces connect into execution graphs.</p><b class="rm-arrow" aria-hidden="true">→</b></a>
<a class="phase" href="roadmap.html#v0-3" data-status="draft"><span class="rm-top"><span class="rm-ver">v0.3</span><span class="rm-status"><i></i>Exploratory</span></span><strong>Semantics</strong><p>Standard event names and lifecycles.</p><b class="rm-arrow" aria-hidden="true">→</b></a>
<a class="phase" href="roadmap.html#v0-4" data-status="draft"><span class="rm-top"><span class="rm-ver">v0.4</span><span class="rm-status"><i></i>Exploratory</span></span><strong>JSON Schemas</strong><p>Artifacts anyone can validate.</p><b class="rm-arrow" aria-hidden="true">→</b></a>
<a class="phase" href="roadmap.html#v0-5" data-status="planned"><span class="rm-top"><span class="rm-ver">v0.5</span><span class="rm-status"><i></i>Planned</span></span><strong>Versioning</strong><p>Rules for safe, compatible changes.</p><b class="rm-arrow" aria-hidden="true">→</b></a>
<a class="phase" href="roadmap.html#v0-6" data-status="planned"><span class="rm-top"><span class="rm-ver">v0.6</span><span class="rm-status"><i></i>Planned</span></span><strong>Extensions</strong><p>Canonical examples and third-party extensions.</p><b class="rm-arrow" aria-hidden="true">→</b></a>
<a class="phase" href="roadmap.html#v0-6x" data-status="planned"><span class="rm-top"><span class="rm-ver">v0.6.x</span><span class="rm-status"><i></i>Planned</span></span><strong>Provenance</strong><p>Evidence, lineage and conformance tests.</p><b class="rm-arrow" aria-hidden="true">→</b></a>
<a class="phase rm-north" href="roadmap.html#v1-0" data-status="north"><span class="rm-top"><span class="rm-ver">v1.0</span><span class="rm-status"><i></i>North star</span></span><strong>Stable standard</strong><p>A frozen spec anyone can implement.</p><b class="rm-arrow" aria-hidden="true">→</b></a>
</div>
</section>
<section class="faq section-shell" id="faq">
<div class="section-heading">
<div><p class="section-label">FAQ / Agentic systems</p><h2>Questions get harder when agents become <span class="hl">autonomous.</span></h2></div>
<p>Explore how DeepAgentLabs helps teams observe, evaluate, govern, and test agentic systems.</p>
</div>
<div class="faq-list">
<article class="faq-item">
<h3 class="faq-heading">
<button class="faq-trigger" id="faq-trigger-01" type="button" aria-expanded="false" aria-controls="faq-answer-01">
<span class="faq-number" aria-hidden="true">01</span>
<span class="faq-question">What happens when AI agents move from demos to production?</span>
<span class="faq-indicator" aria-hidden="true"></span>
</button>
</h3>
<div class="faq-answer" id="faq-answer-01" role="region" aria-labelledby="faq-trigger-01" aria-hidden="true">
<div class="faq-answer-inner">
<p>Production agents need more than model access. They need observability, evaluation, governance, resilience testing, secure access to tools and data, and mechanisms for human intervention. DeepAgentLabs brings these capabilities together through a modular, framework- and runtime-agnostic ecosystem.</p>
<ul class="faq-tags" aria-label="Related capabilities"><li>AgenticOps</li><li>AgenticLens</li><li>Evals</li><li>Sidecar</li><li>Chaos</li><li>MCP</li></ul>
</div>
</div>
</article>
<article class="faq-item">
<h3 class="faq-heading">
<button class="faq-trigger" id="faq-trigger-02" type="button" aria-expanded="false" aria-controls="faq-answer-02">
<span class="faq-number" aria-hidden="true">02</span>
<span class="faq-question">How do you observe what an AI agent is actually doing?</span>
<span class="faq-indicator" aria-hidden="true"></span>
</button>
</h3>
<div class="faq-answer" id="faq-answer-02" role="region" aria-labelledby="faq-trigger-02" aria-hidden="true">
<div class="faq-answer-inner">
<p>AgenticLens provides runtime workflow inspection and analysis. It works with workflow artifacts and traces to surface structured telemetry, recommendations, reports, comparisons, and audit information.</p>
<ul class="faq-tags" aria-label="Related capabilities"><li>AgenticLens</li><li>Observability</li></ul>
</div>
</div>
</article>
<article class="faq-item">
<h3 class="faq-heading">
<button class="faq-trigger" id="faq-trigger-03" type="button" aria-expanded="false" aria-controls="faq-answer-03">
<span class="faq-number" aria-hidden="true">03</span>
<span class="faq-question">How can you know whether an AI agent made the right decision?</span>
<span class="faq-indicator" aria-hidden="true"></span>
</button>
</h3>
<div class="faq-answer" id="faq-answer-03" role="region" aria-labelledby="faq-trigger-03" aria-hidden="true">
<div class="faq-answer-inner">
<p>Agentic evaluation goes beyond checking the final response. Evaluations can score agent behavior, inspect trajectories, benchmark runs, apply LLM-as-Judge methods, run regression tests, and support release-quality decisions.</p>
<ul class="faq-tags" aria-label="Related capabilities"><li>Agentic Evals</li><li>Trajectory Evaluation</li><li>Release Gates</li></ul>
</div>
</div>
</article>
<article class="faq-item">
<h3 class="faq-heading">
<button class="faq-trigger" id="faq-trigger-04" type="button" aria-expanded="false" aria-controls="faq-answer-04">
<span class="faq-number" aria-hidden="true">04</span>
<span class="faq-question">How can teams govern an AI agent's actions?</span>
<span class="faq-indicator" aria-hidden="true"></span>
</button>
</h3>
<div class="faq-answer" id="faq-answer-04" role="region" aria-labelledby="faq-trigger-04" aria-hidden="true">
<div class="faq-answer-inner">
<p>Agentic Sidecar checks proposed actions against declared intent and policies, then allows, warns, or blocks them. This adds a decision-time governance layer without tying the agent to a single framework.</p>
<ul class="faq-tags" aria-label="Related capabilities"><li>Agentic Sidecar</li><li>Governance</li><li>Intent alignment</li></ul>
</div>
</div>
</article>
<article class="faq-item">
<h3 class="faq-heading">
<button class="faq-trigger" id="faq-trigger-05" type="button" aria-expanded="false" aria-controls="faq-answer-05">
<span class="faq-number" aria-hidden="true">05</span>
<span class="faq-question">How can teams test an agent before it reaches production?</span>
<span class="faq-indicator" aria-hidden="true"></span>
</button>
</h3>
<div class="faq-answer" id="faq-answer-05" role="region" aria-labelledby="faq-trigger-05" aria-hidden="true">
<div class="faq-answer-inner">
<p>Agentic Chaos injects controlled faults such as timeouts, rate limits, and tool failures so teams can inspect how workflows respond. Agentic Evals adds repeatable scoring and release gates to help catch regressions before deployment.</p>
<ul class="faq-tags" aria-label="Related capabilities"><li>Agentic Chaos</li><li>Agentic Evals</li><li>Resilience testing</li></ul>
</div>
</div>
</article>
<article class="faq-item">
<h3 class="faq-heading">
<button class="faq-trigger" id="faq-trigger-06" type="button" aria-expanded="false" aria-controls="faq-answer-06">
<span class="faq-number" aria-hidden="true">06</span>
<span class="faq-question">How do the DeepAgentLabs components work together?</span>
<span class="faq-indicator" aria-hidden="true"></span>
</button>
</h3>
<div class="faq-answer" id="faq-answer-06" role="region" aria-labelledby="faq-trigger-06" aria-hidden="true">
<div class="faq-answer-inner">
<p>The ecosystem is designed as modular layers rather than one monolithic runtime. AgenticLens provides observation and analysis, Evals provides behavioral assessment, Sidecar provides supervision and governance, Chaos provides resilience testing, and the MCP server provides a unified interface to these capabilities.</p>
<ul class="faq-tags" aria-label="Related capabilities"><li>Lens → Evals → Sidecar → Chaos → MCP</li></ul>
</div>
</div>
</article>
</div>
</section>
<section class="differentiator section-shell" id="why">
<div class="section-heading">
<div><p class="section-label">Why DeepAgentLabs</p><h2>One contract beats<br><span class="hl">separate tools.</span></h2></div>
<p>Most tools cover a single slice and speak their own format. Here, every stage shares one specification.</p>
</div>
<div class="diff-grid">
<div class="diff-card">
<h3>Observability dashboards</h3>
<p>Show you what already happened. No shared contract with your eval tooling, no CI gate, no fault injection before release.</p>
</div>
<div class="diff-card">
<h3>Framework-native evals</h3>
<p>Locked to whatever agent SDK you started with. A LangGraph run and a CrewAI run can't be compared, gated, or reported the same way.</p>
</div>
<div class="diff-card diff-us">
<h3>DeepAgentLabs</h3>
<p>One spec, seven interoperable tools — observe, stress-test, evaluate, and supervise any framework through the same contract.</p>
</div>
</div>
</section>
<section class="cta cta-join section-shell" id="community" aria-labelledby="community-title">
<div class="cj-copy">
<p class="section-label">Build with us</p>
<h2 id="community-title">Help define how AI systems are operated.</h2>
<p>The standard is early by design. Join the open-source community behind it, or shape the spec directly as a framework author, AI engineer or reliability team.</p>
<ul class="cm-does" aria-label="What the community does"><li><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M3 20h18M5 20V9l7-5 7 5v11M9 20v-6h6v6"/></svg>Workshops</li><li><svg viewBox="0 0 24 24" aria-hidden="true"><rect x="3" y="4" width="18" height="12" rx="1"/><path d="M8 20h8M12 16v4M7 9l2 2-2 2M12 13h4"/></svg>Technical sessions</li><li><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M9 3h6M10 3v6l-5 9a2 2 0 0 0 2 3h10a2 2 0 0 0 2-3l-5-9V3M7.5 15h9"/></svg>Research</li><li><svg viewBox="0 0 24 24" aria-hidden="true"><circle cx="6" cy="6" r="2.5"/><circle cx="6" cy="18" r="2.5"/><circle cx="18" cy="8" r="2.5"/><path d="M6 8.5v7M18 10.5c0 4-6 3-11.2 5.6"/></svg>Open source</li></ul>
<div class="hero-actions">
<a class="button button-light" href="https://github.com/DeepAgentLabs/ai-operations-spec">Join the discussion</a>
<a class="button button-ghost" href="https://github.com/DeepAgentLabs">Follow on GitHub</a>
</div>
</div>
<div class="cm-links">
<a class="cm-link cm-link-main" href="https://discord.gg/fEqPFWgcDG" target="_blank" rel="noopener noreferrer"><span class="cm-link-icon"><i class="fa-brands fa-discord" aria-hidden="true"></i></span><span class="cm-link-text"><strong>Discord</strong><small>Join the server · questions, traces, workshops</small></span><span class="cm-link-go" aria-hidden="true">→</span></a>
<a class="cm-link" href="https://www.linkedin.com/company/deepagentlabs" target="_blank" rel="noopener noreferrer"><span class="cm-link-icon"><i class="fa-brands fa-linkedin-in" aria-hidden="true"></i></span><span class="cm-link-text"><strong>LinkedIn</strong><small>Releases and workshop announcements</small></span><span class="cm-link-go" aria-hidden="true">→</span></a>
<a class="cm-link" href="https://www.instagram.com/deepagentlabs.io" target="_blank" rel="noopener noreferrer"><span class="cm-link-icon"><i class="fa-brands fa-instagram" aria-hidden="true"></i></span><span class="cm-link-text"><strong>Instagram</strong><small>Behind the scenes from sessions</small></span><span class="cm-link-go" aria-hidden="true">→</span></a>
<a class="cm-link" href="https://github.com/DeepAgentLabs" target="_blank" rel="noopener noreferrer"><span class="cm-link-icon"><i class="fa-brands fa-github" aria-hidden="true"></i></span><span class="cm-link-text"><strong>GitHub</strong><small>Star, open issues, send PRs</small></span><span class="cm-link-go" aria-hidden="true">→</span></a>
</div>
</section>
<section class="newsletter section-shell" id="newsletter">
<div class="newsletter-inner">
<div class="newsletter-copy">
<p class="section-label">DeepAgent Labs Newsletter</p>
<h2>Get DeepAgentLabs<br><span class="hl">updates in your inbox.</span></h2>
<p>Release notes, profiling techniques, and early access to new DeepAgentLabs features. No spam, unsubscribe anytime.</p>
</div>
<form action="https://buttondown.com/api/emails/embed-subscribe/deepagentlabs" method="post" target="popupwindow" onsubmit="window.open('https://buttondown.com/deepagentlabs', 'popupwindow')" class="newsletter-form">
<label for="bd-email" class="sr-only">Email address</label>
<input type="email" name="email" id="bd-email" placeholder="you@example.com" required>
<input type="hidden" value="1" name="embed">
<button type="submit" class="button button-primary">Subscribe</button>
</form>
</div>
</section>
</main>
<footer class="site-footer">
<div class="ft-inner">
<div class="ft-about">
<p class="ft-statement">Open infrastructure for <em>observable, efficient and resilient</em> AI agents, built in the open.</p>
<ul class="ft-values" aria-label="Principles"><li>Spec comes first</li><li>Works with any framework</li><li>Runs on your machine</li><li>Free and open source</li></ul>
</div>
<nav class="ft-col" aria-label="Products"><p class="ft-title">Products</p><ul><li><a href="https://github.com/DeepAgentLabs/agenticlens" target="_blank" rel="noopener noreferrer">AgenticLens</a></li><li><a href="https://github.com/DeepAgentLabs/agentic-evals" target="_blank" rel="noopener noreferrer">Agentic Evals</a></li><li><a href="https://github.com/DeepAgentLabs/agentic-sidecar" target="_blank" rel="noopener noreferrer">Agentic Sidecar</a></li><li><a href="https://github.com/DeepAgentLabs/agentic-chaos" target="_blank" rel="noopener noreferrer">Agentic Chaos</a></li><li><a href="https://github.com/DeepAgentLabs/mcp-server" target="_blank" rel="noopener noreferrer">MCP Server</a></li><li><a href="https://github.com/DeepAgentLabs/agenticops-control-tower" target="_blank" rel="noopener noreferrer">Control Tower</a></li></ul></nav>
<nav class="ft-col" aria-label="Explore"><p class="ft-title">Explore</p><ul><li><a href="https://github.com/DeepAgentLabs/ai-operations-spec" target="_blank" rel="noopener noreferrer">Specification</a></li><li><a href="roadmap.html">Roadmap</a></li><li><a href="#architecture">Capability map</a></li><li><a href="#dashboards">Frameworks</a></li><li><a href="#faq">FAQ</a></li><li><a href="https://mcp.deepagentlabs.io" target="_blank" rel="noopener noreferrer">Remote MCP</a></li></ul></nav>
<nav class="ft-col" aria-label="Community"><p class="ft-title">Community</p><ul><li><a href="https://discord.gg/fEqPFWgcDG" target="_blank" rel="noopener noreferrer"><i class="fa-brands fa-discord" aria-hidden="true"></i> Discord</a></li><li><a href="https://www.linkedin.com/company/deepagentlabs" target="_blank" rel="noopener noreferrer"><i class="fa-brands fa-linkedin-in" aria-hidden="true"></i> LinkedIn</a></li><li><a href="https://www.instagram.com/deepagentlabs.io" target="_blank" rel="noopener noreferrer"><i class="fa-brands fa-instagram" aria-hidden="true"></i> Instagram</a></li><li><a href="https://github.com/DeepAgentLabs" target="_blank" rel="noopener noreferrer"><i class="fa-brands fa-github" aria-hidden="true"></i> GitHub</a></li></ul></nav>
</div>
<div class="ft-word" aria-hidden="true"><span class="ft-word-text"><span class="ft-w1">DeepAgent</span><span class="ft-w2">Labs</span></span></div>
<div class="ft-bottom">
<p>© <span id="year">2026</span> DeepAgentLabs</p>
<p class="ft-status"><i aria-hidden="true"></i>AI Operations Specification · v0.1 in review</p>
<a class="ft-top" href="#top">Back to top <span aria-hidden="true">↑</span></a>
</div>
</footer>
</body>
</html>